Verified Behavior¶

This page keeps the strongest repo-backed claims close to the exact commands that reproduce them.

Deterministic Replay¶

Input: - tests/fixtures/tone_cf32.sigmf-data - tests/fixtures/tone_cf32.sigmf-meta

Command:

./build/sdr-analyzer-cli \
  --source replay \
  --input tests/fixtures/tone_cf32.sigmf-data \
  --meta tests/fixtures/tone_cf32.sigmf-meta \
  --frames 4

Observed replay result: - one narrowband detection at 100.15 MHz - strongest peak at -8.87819 dBFS - occupied bandwidth at 5859.375 Hz - replay stops cleanly at end of stream after the fixture is consumed

Notes: - the signal metrics above are stable across repeated runs of the committed fixture - diagnostic timestamps differ across runs, so raw log lines are not byte-identical

Representative CLI output:

Session stopped after 1 frame(s): Source read for source 'replay' failed: Replay file reached end of stream.
frame=0 noise=-140dBFS strongest=-8.87819dBFS detections=1
  peak=1.0015e+08Hz bw=5859.38Hz power=-8.87819dBFS label=narrowband

Example Output¶

The screenshot below was regenerated locally from the committed replay fixture through the Qt GUI in offscreen mode.

Replay-backed FFT example

Hardware Sanity¶

Current repo evidence is intentionally conservative: - replay is fully verified in-repo - rtl_tcp is verified for the code path through a mock transport test - UHD and SoapySDR are implemented, but attached-device validation is still pending

For the current hardware status matrix, see Hardware Validation Status.